Color correction of negatives

[jens_k_olsen]

For a long time I have been checking the forums to get some help with

improving scanning my negatives. I have a Nikon CoolScan ED 5000 and

am using NikonScan. I mainly use Kodak HD400 and 100UC.

 

I have not yet been able to get scans where I am satisfied with the

colors. What ever I do, there always remain some color cast.

 

Basically I understand there there are only 3 sources of the color

cast:

 

1: The orange mask of the negative

2: The color temperature at the time of exposure

3: Color cast from the scanner itself.

 

In my case the color cast mainly seems to be from

NikonScan's "interpretation" of the orange mask. Since both Vuescan

and SilverFast are using some preset data to remove the orange mask,

I have already tried these programs also to see if that would solve

my problem. Unfortunately none of the programs have specific data

sets for the film I am using, and maybe therefore, I was not

successful.

 

My next try was to scan as positive and then subtract the orange mask

myself and do the conversion to a positive image. I scanned an

unexposed negative to find out what RGB values would make the orange

mask white, and then I adjusted the analog gain accordingly before

scanning. After I inverted the image, I found that it still had a

little light blue cast, and the image was really flat.

 

The method I am using now is to scan a negative at autoexposure,

which does not seem to have any color cast, and then turn off

autoexposure for the following scans. This method has so far given me

the best results, but there is still a greenish color cast.

 

Is it really not possible to perfectly balance out the orange mask?

Kodak certainly must know everything about the orange mask in their

films, so if this information is available, it should be possible to

adjust a scanner to perfectly balance out the orange mask.

 

When I get my small prints developed by Kodak here in Japan, the

colors are excellent. I can not see any color cast.

 

Any advise is highly appreciated (other than use Vuescan or

SilverFast, without specifically explaining why).

 

Thanks a lot !

[jens_k_olsen]

This is the photo I usually use to "calibrate" the scanner before I scan. It's my youngest son sleeping on the bicycle.

[jens_k_olsen]

This is the next frame on the negative strip

[marcos_rodriguez]

Jens, first of all, I feel your pain. Like the saying goes, been there, done that.... I have a Minolta Scan Elite 5400 and, like your case with Nikon's software, I've been disappointed with its results on negatives --mainly, excessive clipping of shadows and highlights.

 

After much pain and experimentation (see this thread: http://www.photo.net/bboard/q-and-a-fetch-msg?msg_id=00CWmS) I've concluded that the best method is:

 

1. Scan "raw" with the scanner's software (in the Minolta, the option is called "16Bit Linear)

 

2. Convert to positive using SilverFast HDRStudio.

 

3. Use Photoshop's Curves's "Autocolor" function to remove remaining color cast and improve contrast.

 

Using the above method, I've achieved colorcast free images with good color balance and saturation. Of course, considering all the pain involved in the process, I'm leaving film for good and going digital (Canon 350D)!

 

Sadly, I still have an extensive negative archive that needs to be scanned, so....

 

Best of luck,

 

Marcos Rodriguez / www.aukeramen.com

[marcos_rodriguez]

Have you tried running the images through Photoshop's autocolor?:

 

Curves > Options > Enhance Per Channel Contrast, check "Snap Neutral Midtones"

 

I did it with your second image and it helped a lot. I suspect that using the original, uncompressed files should give far better results.

 

Best,

 

Marcos

[jens_k_olsen]

Thanks for your advise Marcos, but my whole point is that it must be possible to correct for at least the orange mask during scanning.

 

I have searched the Kodak home page and found something called Film Terms. I downloaded some of them and looked at them. There is quite a lot of information in there, the problem is just, how to use it if you are not using Kodak equipment. I was hoping that somebody with knowledge of these film terms and how they are being used could shed some light on it.

[alan_chan4]

It might not solve your problem entirely, but I use Minolta Elite with Vuescan. My work flow is like this:

 

1) load a blank frame of the roll you are going to scan

2) preview

3) lock exposure

4) preview

5) lock film base color

6) black & white point = 0

 

The negative type I used for HD400 & 400UC are:

Kodak > Royal > Gold400Gen2

 

Not sure about HD200 & 100UC, but you can also play with "Kodak > Supra" for these films. You might read more here: http://www.tedsimages.com/text/comment5.htm

[bruce_rubenstein]

It is an over simplification to think that all that the scanner software is accounting for the orange base; it isn't. The algorithims try to arrive at a neutral color balance, but it can't know the intent or subject in the photograph so the color balance is generally a bit off. This is no different than when you get machine prints done from film. Unless there is an operator monitoring the machine's interpertation of proper color balance, it's also usually off.

 

I have the Nikon Coolscan V, HP20s and had a Canon FS4000. None of them are, or were, close to perfect. The HP software is the best for doing prescan adjustments. Nikon Scan sucks because you have to do a slow preview before you can see what your adjustments actually do. With the Nikon, I look at the thumbnails and if the colors/exposures are off quite a bit, I'll scan in 16 bits, otherwise 8 bits. Final exposure/color correction is done in PS. There's just no way around it.

[mendel_leisk]

"My next try was to scan as positive and then subtract the orange mask myself and do the conversion to a positive image. I scanned an unexposed negative to find out what RGB values would make the orange mask white, and then I adjusted the analog gain accordingly before scanning. After I inverted the image, I found that it still had a little light blue cast, and the image was really flat."

 

This is precisely what I've done, and I'm getting the same results.

 

I would think the image is really flat because that's the nature of color negative film: it is able to capture such a wide range of tones. With adjustment through Photoshop's levels and curves, you should be able to get to a more pleasing contrast.

 

The cyan cast continues to puzzle me. If analogue gain adjustments are yielding white scans of the leader, why are frames with content having cyan cast? Like Marcos, I resorted to Photoshop Autocolor in my workflow, except I found "snap neutral midtones" was taking the color balance off in new and undesired directions, as often as not.

 

For me, using a wider color space, say Adobe RGB instead of sRGB, made a big improvement in Photoshop's Autocolor results. Images where nothing seemed to eliminate the cast, suddenly were close to neutral. A clue?

 

I think you're likely on the right track, researching Kodak "terms", but am clueless.

 

I suspect if and when you do hit on something that neutralizes your scans, it will be anti-climatic. Color cast is obvious, tangible. Balanced color, on the other hand, doesn't attract attention. Not exactly sure what I'm saying, hope that makes some sense.

 

What puzzles me is that this can be such a big mystery. There is such a massive amount of color negative film out there, and so many people trying to digitize their collections by scanning. All the folks that have "cracked" it are keeping quiet?

 

I've "shelved" it for now, having a massive backlog of slides to scan, and a 20D to play with.

 

With your interest, you've more than likely found this already. Marcos really carries the ball here. I don't know that anything is totally resolved, but anyway:

 

http://www.photo.net/bboard/q-and-a-fetch-msg?msg_id=00CWmS

 

And this should also be of interest:

 

http://groups-beta.google.com/group/comp.periphs.scanners/browse_frm/thread/42c3b2fa6068fc62?tvc=1&q=elite+5400+film+strip+holder&hl=en

[Ed_Ingold]

You can develop a custom profile for negative film using a Gretag-Macbeth Color Checker chart in the image and "InCamera" plugin for Photoshop by PictoColor.

 

You have to create several profiles for each film type, depending on the lighting, but "daylight", "cloudy", "shade" and "tungsten" would probably get you close enough.

 

These profiles are used by "assigning" (not "converting") the profile to the image in Photoshop, then "converting" to a standard color space like Adobe RGB.

 

SilverFast can accumulate adjustments which can be hard to find and remove. Sometimes it's best to reset the settings (alt-options) and start over. In general, use the least possible adjustment and scan in 14-bit mode.

 

"Autocolor" or "Levels/Options" in Photoshop is a quick but crude way to achieve color balance. If you find a setting in "Levels" that seems to work for you, save it by name. "Curves" can be used for fine-tuning and gamma control.

[david_moore13]

Removing the orange mask correctly is not that hard to do, so I suspect most software does a decent job with it. However, color casts can come from many other sources because film inversion is such a complicated procedure. Basically, the colors in the image get dragged through numerous different "color spaces" before you are done with the image:

 

1. The red, green, and blue (and maybe cyan too) spectral sensitivity of the film.

 

2. The cyan, magenta, yellow (and maybe red) color of the film's dye layers.

 

3. The red, green, and blue spectral sensitivity of the film scanner and its light source.

 

4. The working color space of the software.

 

5. The color of your monitor's phosphors.

 

Since 2-3 are imaging the negative rather than the positive, there's a good chance that all these color space conversions will result in some loss of color information. That's why color negatives are so painful to scan. Some types of film work better than others because color spaces 1-5 are a closer match to each other.

 

Many of these problems go away when you print photographic paper because you no longer have to deal with color spaces 3-5. Instead you work with full-spectrum light.

 

Also, film makers strive to make sure that the characteristic curves for red, green, and blue are all identical (except for a constant density shift between each -- the mask color). However, whenever the characteristic curves have slightly different slope, or slightly different sensitivity to light, you will probably get a color cast in your output. This is greatly exacerbated if you use a colored filter over the lens.

[jtk]

You SHOULD be able to print very accurate color (not speaking here of shadow/highlight detail) simply by using your scanner's film default settings and your printers paper defaults.

 

Use the most basic instructions that came with your scanner and

printing software.

 

In my experience Epson, Silverfast, Nikon and Minolta all provide excellent defaults with excellent software for printing color negatives.

 

It's true that there are subtle differences between different color negative stocks, but they become insignificant if you boost your contrast just a little. Yes, you can install special profiles for each type of color negative, but that's not the answer to your "color cast" question.

[bobmichaels]

I've been scanning film for some 4-5 years. My scanner is calibrated with an IT8 target. My monitor is calibrated with an Optix XR. I have custom profiles for my printer, paper, ink combinations. I believe I understand the basics of color management.

 

I have yet to make a perfect print without some adjustment in Photoshop.

[albert lui]

Photoshop's tools are very convenient, so I just try to obtain a good scan (full tonal range) in 16 bit color. Then I postprocess in PS for color correction. Barry Hayne's Photoshop Artistry book (one for each version of PS) is a great tutorial for postprocessing.